Morphologically Unbiased Classifier Combination through Graphical PDF Correlation

  • David Windridge
  • Josef Kittler
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 2396)


We reinterpret the morphologically unbiased’ tomographic’ method of multiple classifier combination developed previously by the authors as a methodology for graphical PDF correlation. That is, the original procedure for eliminating what are effectively the back-projection artifacts implicit in any linear feature-space combination regime is shown to be replicable by a piecewise morphology matching process. Implementing this alternative methodology computationally permits a several orders-of-magnitude reduction in the complexity of the problem, such that the method falls within practical feasibility even for very high dimensionality problems, as well as resulting in a more intuitive description of the process in graphical terms.


Probability Density Function Cycle Count Computational Implementation Reconstructive Space Pattern Recognition Letter 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


  1. 1.
    R. A. Jacobs, “Methods for combining experts’ probability assessments”, Neural Computation, 3, pp. 79–87, 1991CrossRefGoogle Scholar
  2. 2.
    J. Kittler, M. Hatef, R. P. W. Duin, and J. Matas, “On combining classifiers”, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 20, no. 3, 1998, 226–239CrossRefGoogle Scholar
  3. 3.
    L. Lam and C. Y. Suen, “Optimal combinations of pattern classifiers”, Pattern Recognition Letters, vol. 16, no. 9, 1995, 945–954.CrossRefGoogle Scholar
  4. 4.
    A. F. R. Rahman and M C Fairhurst, “An evaluation of multi-expert configurations for the recognition of handwritten numerals”, Pattern Recognition Letters, 31, pp. 1255–1273, 1998Google Scholar
  5. 5.
    A. F. R. Rahman and M C Fairhurst, “A new hybrid approach in combining multiple experts to recognise handwritten numerals”, Pattern Recognition Letters, 18, pp. 781–790, 1997CrossRefGoogle Scholar
  6. 6.
    K. Woods, W. P. Kegelmeyer and K Bowyer, “Combination of multiple classi-fiers using local accuracy estimates”, IEEE Trans. Pattern Analysis and Machine Intelligence, 19, pp. 405–410, 1997CrossRefGoogle Scholar
  7. 7.
    D. Windridge, J. Kittler, “An Optimal Strategy for Classifier Combination: Part 1: Multiple Expert Fusion as a Tomographic Process”, (PAMI, Submitted)Google Scholar
  8. 8.
    D. Windridge, J. Kittler, “An Optimal Strategy for Classifier Combination: Part 2: General Application of the Tomographic Procedure”, (PAMI, Submitted)Google Scholar
  9. 9.
    D. Windridge, J. Kittler, “Classifier Combination as a Tomographic Process”, (Multiple Classifier Systems, LNCS. Vol. 2096, 2001.)Google Scholar
  10. 10.
    D. Windridge, J. Kittler, “A Generalised Solution to the Problem of Multiple Expert Fusion.”, (Univ. of Surrey Technical Report: VSSP-TR-5/2000)Google Scholar
  11. 11.
    F. Natterer, Proceedings “State of the Art in Numerical Analysis”, York, April 1–4, 1996.Google Scholar
  12. 12.
    J. Högbom, “Aperture synthesis with a non-regular distribution of interferometer baselines”, Astrophys. J. Suppl. Ser., 15, 417–426, 1974Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2002

Authors and Affiliations

  • David Windridge
    • 1
  • Josef Kittler
    • 1
  1. 1.Centre for Vision, Speech and Signal Processing Dept. of Electronic & Electrical EngineeringUniversity of SurreyGuildfordUK

Personalised recommendations